Silicon Investor (SI) -- The First Internet Community

STOCKTALK

We've detected that you're using an ad content blocking browser plug-in or feature. Ads provide a critical source of revenue to the continued operation of Silicon Investor. We ask that you disable ad blocking while on Silicon Investor in the best interests of our community. If you are not using an ad blocker but are still receiving this message, make sure your browser's tracking protection is set to the 'standard' level.

Technology Stocks : Advanced Micro Devices - Moderated (AMD) -- Ignore unavailable to you. Want to Upgrade?

To: combjelly who wrote (239609)	8/30/2007 6:31:25 PM
From: dougSF30	Read Replies (1) \| Respond to of 275872

Obviously, Intel thinks it is best to do what they've done. :)

To: combjelly who wrote (239609)	8/30/2007 7:05:43 PM
From: graphicsguru	Read Replies (2) \| Respond to of 275872

What I am saying is that such cases in a large NUMA system are going to be really slow . . . Certainly the unwind cost goes up in a large NUMA system, but if only the affected cores are involved in the unwind, the unwinding time penalty wouldn't explode. Of course, any unwind buffers might have to be larger. Maybe that's part of why the MP versions of CSI are a year later. In fact the speculation should yield a bigger payoff for large NUMA systems. Opteron already loses almost a speed grade when going from 2S to 4S because of the double-hop coherency latency. The more hops involved in coherency, the more you can win by speculating when you're right.